High Pressure Research
Vol. 26, No. 4, December 2006, 539–547
Static stress demagnetization of single and multidomain
magnetite with implications for meteorite impacts
STUART A. GILDER*†‡, MAXIME LE GOFF† and JEAN-CLAUDE CHERVIN§
†Institut de Physique du Globe de Paris, Equipe de Paléomagnétisme, 4 place Jussieu,
75252 Paris Cedex 05, France
‡Department of Earth and Environmental Sciences, Geophysics Section, Ludwig Maximilians
University, Theresienstrasse 41, 80333 Munich, Germany
§Département de Physique des Milieux Denses, Institut de Minéralogie et de Physique des Milieux
Condensés, Université Pierre et Marie Curie, CNRS-UMR 7590, 140 rue de Lourmel,
75015 Paris, France
(Received 3 July 2006; revised 25 August 2006; in final form 22 September 2006)
Stress demagnetization effects on ferromagnetic minerals are poorly known, especially above 1 GPa,
and when initially magnetized under pressure and then subjected to further stress. Our experiments
on pure magnetite under quasi-hydrostatic loads in the presence of a small (Earth’s) field show that
stress demagnetization depends on domain state and stress history. Viewed globally, the results follow
a simple law where the percentage loss in magnetic moment is the inverse of pressure (e.g., 50% loss in
moment at 1 GPa, 67% loss at 2 GPa, etc.). Our experiments also quantify the effect of demagnetization
upon stress release, where the moment upon full decompression is two-thirds less than the moment
when decompression first began. Given the magnitude of the stress demagnetization effect, we conclude
that the presence or absence of a planetary magnetic field cannot be deduced from the magnetic fields
measured over meteorite craters, such as those on Mars.
Keywords: Stress; Demagnetization; Magnetite; Meteorite crater
1.
Introduction
With the advent of paleomagnetic studies in the 1940s, scientists began to wonder how the
effect of stress, from burial, folding, etc., would influence the magnetic remanence of rocks.
By the mid 1950s, an intense research effort was underway aimed at understanding piezoremanence, which is the remanent magnetization produced by stress, as well as the opposite
effect of stress demagnetization. Laboratory experiments suggested that the magnetic signals of
rocks were sensitive to stresses typically found around fault zones [1–5]. And it was calculated
*Corresponding author. Tel.: +33-144274934; Email: [email protected] and [email protected]
High Pressure Research
ISSN 0895-7959 print/ISSN 1477-2299 online © 2006 Taylor & Francis
http://www.tandf.co.uk/journals
DOI: 10.1080/08957950601092085
540
S. A. Gilder et al.
that such stress-induced changes would in turn modify the local magnetic field [6–9]. Scientists
thought that by monitoring magnetic field variations, earthquakes or volcanic eruptions could
be predicted [10]. However, field observations of a pressure-induced magnetic effect proved
elusive. More convincing evidence for stress-induced magnetization changes in nature comes
from meteorite impact craters, which are often characterized by lower than average magnetic
anomalies. For example, magnetic surveys of the Martian surface detect significantly lower
magnetic field intensities over the gigantic impact craters Hellas and Argyre than over the
surrounding regions [11–13]. The reduced fields are commonly attributed to pressure demagnetization caused by shock waves generated during meteorite impact in the absence of a
planetary field [13–15].
The effect of stress in the absence of an external magnetic field has a permanent demagnetizing effect on rocks that contain an assemblage of magnetic grains. In this case, permanent
means that once the pressure is released, the magnetic moment does not grow back to
its original (pre-compressed) value because the magnetic vector directions of the grains
have been randomized, thus lowering the net magnetization of the rock. It does not mean
that the individual magnetic grains have lost their capacity to be magnets. This makes
stress demagnetization similar in theory to alternating field demagnetization [16]. However,
quantifying stress demagnetization is difficult because the magnetization of the pressure
vessel must be significantly lower than that of the sample, and the cell must not interfere
with the electronics of the detector. In order to amplify the magnetic signal, one needs
to maximize the number of magnetic grains, which explains why most experimental work
on this problem has focused on large rock samples. Because a trade-off exists between
pressure and sample size, most experiments have been performed using uniaxial stresses,
at pressures lower than 1 GPa (10 kbar), with most measurements taking place after pressure
release.
Despite the experimental restrictions, several aspects of stress demagnetization are known.
Martin and Noel [17] demonstrated the differences between hydrostatic and uniaxial stress
demagnetization on a diabase rock containing large (20–300 micron) titanomagnetite (Curie
temperature = 535 ◦ C) grains. Their experiments showed that uniaxial stresses demagnetize
this material about two times more than hydrostatic stresses; e.g., at 100 MPa, hydrostatic
and uniaxial pressures demagnetize roughly 20% and 40% of the original, non-compressed
magnetization intensity, respectively. Adding a shear component to a static uniaxial stress
will demagnetize rocks even faster [18]. Another observation from high-pressure experiments performed under uniaxial stress is that magnetic properties change in relation to
the maximum stress axis direction [19–22]. For stress demagnetization, the magnetization
intensity decreases faster parallel to the maximum stress direction rather than perpendicular to it; magnetic susceptibility increases perpendicular to the maximum stress direction
and decreases parallel to it. Kletetschka et al. [23] shocked a variety of magnetic mineral
aggregates to pressures around 1 GPa. They found a correlation between the percentage of
retained magnetization and magnetic coercivity, consistent with previous work [e.g., 24–27].
Gattacceca et al. [28] combined pulsed laser shocks with magnetic microscopy on natural
titanomagnetite-bearing basalt. They found that 1 GPa pressures remove 70–80% of the initial
magnetization.
To our knowledge, previous investigations have largely ignored how grain size, composition,
etc. bear on stress demagnetization while the sample is under pressure. Moreover, decompression effects have not been quantified. For these reasons, we developed new high-pressure cells
that facilitate direct measurements of the full magnetic vector to pressures of ∼40 GPa, which
allows one to experiment on magnetic minerals whose composition and grain sizes are well
characterized.
Static stress demagnetization
2.
541
Experimental procedure and results
We designed a Merrill–Basset-type diamond anvil cell made solely of BeCu, including the
aligning screws. Fe-pure (Ti-free) single and multidomain magnetite, previously described in
Gilder et al. [29, 30], were loaded together with ruby spheres and silica gel in a cylindrical
chamber drilled in a bronze-Be gasket with a diameter and height (after work hardening) of
400 μm and 100 μm, respectively. Silica gel served as a pressure medium to maintain hydrostatic pressures, as attested to by well-defined R1 and R2 peaks in the ruby florescence spectra
[31]. Stresses were not perfectly hydrostatic, as a 10% difference in pressure exists between
rubies in the center of the cell versus those near the edge. The grains easily disaggregate
after depressurizing from 5 GPa; they are not compacted into a solid mass. After loading, a
direct current (dc) field of 315 mT was applied parallel to the plane of the diamond cutlets so
that the samples would acquire an isothermal remanent magnetization (IRM). The magnetic
remanence of the sample was measured by inserting the cell directly into a 2G Inc. three-axis
cryogenic magnetometer. Pressure was then raised or lowered sequentially in the Earth’s magnetic field in Paris (∼48,000 nT) at room temperature, with the magnetic remanence being
measured at each pressure step. We also applied a 315 mT field to the samples while they were
under pressure to observe the stress demagnetization effect on a sample whose magnetization was acquired under stress. The cell itself acquires a slight magnetic remanence from the
applied field, which remains constant throughout an experiment, that was subtracted from the
total measured moment.
Figure 1 shows the pressure path and corresponding magnetic moments of the experiments
performed on the same multidomain magnetite sample (data in table 1). Starting from ambient
pressure (P = 0 GPa), the magnetization decreased to 21% of the original value by 2.03 GPa.
After a new 315 mT field was imposed at 2.03 GPa, the moment decreased faster with increasing pressure than when starting from ambient conditions. This suggests that a multidomain
Figure 1. Stress demagnetization of multidomain magnetite. A direct field of 315 mT was applied to the sample at
ambient conditions to create an isothermal remanent magnetization (IRM) (‘start’ in figure). The magnetic remanence
was then measured under pressure at successively higher pressures until 2.03 GPa, then a new 315 mT field was applied
to the sample, which was then compressed to successively higher pressures until 4.23 GPa. Pressure was then released,
a new 315 mT field imposed, followed by a third round of stress demagnetization.
542
S. A. Gilder et al.
Table 1.
Stress demagnetization data.
Single domain magnetite
Pressure (GPa)
M
(Am2
IRM – 1st compression
0.16
0.38
0.82
1.07
1.57
1.81
× 10−6 )
Multidomain magnetite
M norm
0.365
0.216
0.150
0.119
0.089
0.054
1.00
0.59
0.41
0.33
0.24
0.15
IRM – 2nd compression
1.81
1.132
1.92
1.062
2.11
0.952
1.00
0.94
0.84
1st decompression
2.11
1.48
1.02
0.60
0.16
0.952
0.861
0.742
0.412
0.315
1.00
0.90
0.78
0.43
0.33
IRM – 3rd compression
0.16
0.800
0.85
0.672
1.29
0.571
1.95
0.509
2.22
0.491
1.00
0.84
0.71
0.64
0.61
IRM – 2nd decompression
2.22
1.212
1.57
1.112
1.21
0.942
0.93
0.766
0.55
0.476
0.11
0.395
1.00
0.92
0.78
0.63
0.39
0.33
Pressure (GPa)
M (Am2 × 10−6 )
IRM – 1st compression
0.00
0.66
1.40
1.68
2.03
0.163
0.125
0.055
0.043
0.034
M norm
1.00
0.77
0.34
0.26
0.21
IRM – 2nd compression then decompression
2.03
0.225
1.00
2.31
0.127
0.56
2.50
0.093
0.41
2.77
3.27
4.23
0.38
0.066
0.044
0.036
0.013
0.29
0.20
0.16
0.06
IRM – 3rd compression
0.38
0.700
0.77
0.592
1.26
0.447
2.03
0.244
3.13
0.169
1.00
0.85
0.64
0.35
0.24
Abbreviations: M: magnetic moment; M norm: magnetic moment normalized by the maximum value in a given run; IRM:
isothermal remanent magnetization. Non-normalized data are shown in figures 1 and 2.
magnetite-bearing rock that acquired its moment at depth, and was subsequently compressed,
demagnetizes faster than an equivalent rock that formed near the surface and was then compressed. Although the decompression effect on the moment appears negligible in figure 1,
the moment actually decreases nearly three times from 4.23 GPa to 0.38 GPa. Upon a new
round of compression, the moment decreased, but slower than during the first compression
run (figure 1).
Figure 2 shows the pressure path and corresponding magnetic moments of the experiments
performed on the same single domain magnetite sample (data in table 1). We paid close
attention to the effects of decompression in these experiments. Starting from ambient pressure,
the magnetization decreased to 15% of the original value by 1.81 GPa. After a new 315 mT
field was applied at 1.81 GPa, the IRM moment decreased slower with increasing pressure
than when starting from ambient conditions. This suggests that a single domain magnetitebearing rock that acquired its remanence at depth and was then compressed demagnetizes
slower than an equivalent rock formed at the surface and was then compressed. The sample
was then stepwise decompressed to 0.16 GPa, a 315 mT field was imparted, followed by a
third round of compression up to 2.22 GPa. Another 315 mT field was applied at 2.22 GPa,
then the pressure was incrementally decreased.
Static stress demagnetization
543
Figure 2. Stress demagnetization of single domain magnetite. A direct field of 315 mT was applied to the sample at
ambient conditions to create an isothermal remanent magnetization (IRM) (‘start’ in figure). The magnetic remanence
was then measured under pressure at successively higher-pressures until 1.81 GPa, then a new 315 mT field was applied
to the sample. This newly acquired IRM moment was then measured under pressure at successively higher-pressures
until 2.11 GPa, and then pressure was sequentially released. A new 315 mT field was imposed, followed by a new round
of stress demagnetization until 2.22 GPa, where another 315 mT field was applied, and then the cell was sequentially
decompressed.
We normalized the magnetic moment during a given compression or decompression run to
the moment measured at the start of the run (the IRM moment in most cases) (figures 3 and 4).
For the compression runs (figure 3), pressure is displayed relative to the initial pressure at the
beginning of the run, called delta pressure (P ), which is the pressure at step n (Pn ) minus
the initial pressure (Pi ) when the particular run began (P = Pn − Pi ). For decompression
(figure 4), pressure is normalized relative to the pressure when decompression began (Pn /Pi ).
Qualitatively, for the first compression cycle (figure 3), which might correspond to a rock that
crystallized near the surface and was then compressed, one sees that multidomain magnetite
demagnetizes less easily than single domain magnetite. For magnetite that cooled through its
Curie point while under pressure, such as a rock lying at considerable depth in a slowly cooling
planet, the opposite is true, single domain magnetite better resists pressure demagnetization
than multidomain magnetite. Finally, for work-hardened minerals, e.g., those subjected to
multiple stress cycles, such as a rock bombarded by successive impacts yet somehow remagnetized after the last impact, both single and multidomain magnetite increase their resistance
to pressure demagnetization.
The results follow fairly simple quantitative rules. Previously uncompressed single
domain magnetite or multidomain magnetite formed at depth will lose their magnetizations
roughly as MP n /MP i = 1/Pn2 (or approximately exponentially MP n /MP i = e−P n ), where
(MP n /MP i ) × 100 is the percentage of the initial moment (MP i ) remaining at Pn , which is
the pressure, in GPa, above the initial pressure, Pi (where Pi is treated in the same way as in
figure 3). Previously unstrained multidomain magnetite or single domain magnetite-bearing
rocks formed at depth will lose their remanences roughly following a linear law valid below
1.5 GPa where MP n /MP i ≈ −0.45Pn + 1. In the case of large meteorite impacts where pressure waves penetrate deeply (several 10s of kilometers), one could use a general rule of 1/P
for the decrease in the percentage of original magnetization for the entire volume of affected
544
S. A. Gilder et al.
Figure 3. Synthesis of pressure demagnetization results. Single domain (SD) and multidomain (MD) grains originally lying close to the surface will loose half their initial remanence by 0.5 and 1.2 GPa, respectively (open symbols).
Grains that formed at depth (filled symbols), then subject to further compression, show nearly the opposite behavior,
with MD grains demagnetizing faster than SD grains. For stress-cycled grains, e.g., those previously subjected to
stress and then released from stress, then recompressed, multidomain grains show essentially the same behavior as
in the initial run, whereas SD magnetite becomes much more resistant (gray symbols). Delta pressure (P ) is the
pressure at step n(Pn ) minus the initial pressure (Pi ) when the particular run began (P = Pn − Pi ).
Figure 4. Stress demagnetization of single domain (SD) magnetite under decompression. Two separate runs on the
same SD sample yield quite compatible results. Data points at the beginning and end of demagnetization are also
shown for multidomain (MD) magnetite. The total relative effect appears to be the same regardless of grain size or
starting pressure.
Static stress demagnetization
545
material, although it would be a slight underestimate at higher pressures. We can also propose
a specific law that fits each demagnetization curve as
MP n /MP i = fnon-demag + (1 − fnon-demag ) × exp(−(Pn /Pc )),
where fnon-demag is the fraction of the moment incapable of being demagnetized even at the
highest imposed pressures (0.15 seems to best fit the data – figure 3), and Pc is a characteristic
pressure parameter which varies from run to run. One varies Pc in order to fit each run; for
example, a Pc of 0.4 and 3.0 matches well the first demagnetization run for multidomain
magnetite and the third compression run for single domain magnetite, respectively.
The decompression results follow the same trend regardless of the compression path or
number of compressions (figure 4). From a maximum initial pressure (Pi ), the moment will
decrease as MP n /MP i ≈ 0.8 × Pn /Pi + 0.3 [or MP n /MP i ≈ 0.45 × ln(Pn /Pi ) + 1, when
Pn > 30% of Pi ]. A simplified representation that includes all the decompression data is a
law of MP n /MP i = (Pn /Pi )0.5 , with a general rule being that two-thirds of the remanence
is lost between the point when compression stops and the end of full decompression (Pn ≈
0 GPa). In an absolute sense, demagnetization during pressure release is minor for samples
having undergone significant compression (>1 GPa). However, if a rock is exhumed from
great depth, as in the central dome region of large craters, decompression demagnetization
could be important.
3.
Discussion and conclusions
To apply pressure demagnetization in nature, such as on Mars, one needs to know which magnetic phases are being demagnetized. Although pyrrhotite occurs in SNC meteorites, to date
only iron oxides (magnetite, maghemite and hematite) have been identified on the Martian
surface [32], with magnetite being the only magnetic phase found in igneous outcrops. Moreover, using rock magnetism and thermal modeling, Dunlop and Arkani-Hamed [33] evaluated
the candidate minerals responsible for the strong magnetic anomalies of Terra Sirenum and
Terra Cimmeria in Mars’ southern highlands. The prime candidates, in order of likelihood, are
single-domain magnetite, single-domain pyrrhotite and either single or multidomain domain
hematite. Thus both direct observation and theoretical considerations suggest magnetite is the
most likely magnetic mineral in Martian rocks. If true, then our results confer best with the
pressure model of Mohit and Arkani-Hamed [15] for the large Martian craters, as much larger
regions should be demagnetized than what is observed when considering other models.
One potential pitfall of our results, as with most of the existing studies [23, 28], is that an
isothermal remanent magnetization (IRM), and not a thermal remanent magnetization (TRM),
was demagnetized. Assuming that an IRM is more easily demagnetized than a TRM, both our
results and the existing ones may be underestimated. On the other hand, shock demagnetization
experiments by Pohl et al. [25] suggest that samples initially possessing either an IRM or a
TRM yield similar results. Our results differ from others in that single domain magnetite
demagnetizes faster than multidomain magnetite during the first compression runs. This could
be due to the fact that, in our experiments, the dc field was applied perpendicular to the
maximum stress direction. As stress reorients Bloch walls [34, 35], and as the reorientation
process differs at different pressure intervals [36], the differences are likely linked to the
way the domains reorient in relation to the nature of the stress and the angle between the
magnetization and applied stress directions. If an IRM is easier to demagnetize than a TRM,
the effect is potentially cancelled out by the fact that magnetic moments oriented perpendicular
to the maximum stress axis are harder to demagnetize. Another question is whether dynamic
546
S. A. Gilder et al.
(shock) and static demagnetization are comparable. Shock experiments currently underway
suggest this may not be the case for pyrrhotite, but the effects of shock are similar to those
found in static experiments on magnetite [37].
With the above-mentioned caveats in mind, our results bear on the way the magnetic signatures of meteorite craters are used to quantify details surrounding impact events. The volume
of demagnetized rock, and the level of demagnetization, and the size of the crater, are the key
data required to constrain the pressure gradients stemming from the shock [15, 38, 39]. Such
knowledge is needed to ascertain the size of the impactor, its impact velocity, impact angle, etc.
For the large (300–500 km diameter) Martian meteorite impacts, Mohit and Arkani-Hamed
[15] calculated that pressures exceeding 1 GPa extend to about one crater radii away from
the center. Other models suggest that the 1 GPa isobar continues significantly farther than
two basin radii from the Martian craters [13, 40], and two to four crater radii for the lunar
craters [14] (see also refs. [41, 42] for discussion). Another important finding is that, even in
the presence of a small (Earth’s) magnetic field, both compression and decompression have a
demagnetizing effect. These results lead to the important conclusion that one cannot deduce
whether a magnetic field existed on a planet based on the magnetic signatures of the rocks in
meteorite impact craters at the time of impact.
Acknowledgements
We thank Ales Kapicka and two anonymous reviewers for their helpful suggestions. This is
IPGP contribution No. 2171.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
T. Nagata and H. Kinoshita, Nature 204 1183 (1964).
T. Nagata and H. Kinoshita, J. Geomagn. Geoelectr. 17 121 (1965).
T. Nagata, J. Geomagn. Geoelectr. 18 73 (1966).
J.P. Pozzi, Effets de Pression en Magnétisme des Roches. Ph.D. thesis, Université Paris VI, Paris, France (1973).
Y. Hamano, J. Geomagn. Geoelectr. 35 155 (1983).
J.W. Kern, J. Geophys. Res. 66 3807 (1961).
F.D. Stacey, Phil. Mag. 7 551 (1962).
F.D. Stacey, Pure Appl. Geophys. 58 5 (1964).
J. Zlotnicki and F.H. Cornet, J. Geophys. Res. 91 709 (1986).
B.E. Smith and M.J.S. Johnston, J. Geophys. Res. 81, 3556 (1976).
M.H. Acuña, J.E.P. Connerney, P. Wasilewski, et al., J. Geophys. Res., E 106 23403 (2001).
F. Nimmo and M.S. Gilmore, J. Geophys. Res. E 106 12315 (2001).
L.L. Hood, N.C. Richmond, E. Pierazzo, et al., Geophys. Res. Lett. 30 1281 doi:10.1029/2002GL016657
(2003).
J.S. Halekas, D.L. Mitchell, R.P. Lin, et al., Geophys. Res. Lett. 29 1645 doi:10.1029/2001GL013924 (2002).
P.S. Mohit and J. Arkani-Hamed, Icarus 168 305 (2004).
D.J. Dunlop, M. Ozima and H. Kinoshita, J. Geomagn. Geoelectr. 21 513 (1969).
R. Martin and J. Noel, Geophys. Res. Lett. 15 507 (1988).
K.A. Valeev and S.S. Absalyamov, Izv. Phys. Solid Earth 36 241 (2000).
W. Kean, R. Day, M. Fuller, et al., J. Geophys. Res. 81 861 (1976).
M. Lanham and M. Fuller, Geophys. Res. Lett. 15 511 (1988).
A. Kapicka, J. Geophys. 53 144 (1983).
A. Kapicka, Phys. Earth Planet. Int. 63 78 (1990).
G. Kletetschka, J.E.P. Connerney, N.F. Ness, et al., Meteorit. Planet. Sci. 39 1 (2004).
R.S. Carmichael, J. Geophys. 34 775 (1969).
J. Pohl, U. Bleil and U. Hornemann, J. Geophys. 41 23 (1975).
S.M. Cisowski and M. Fuller, J. Geophys. Res. 83 3441 (1978).
G.J. Borradaile and M. Jackson, Phys. Earth Planet. Int. 77 315 (1993).
J. Gattacceca, M. Boustie, B.P. Weiss, et al., Geology 34 333 (2006).
S.A. Gilder, M. Le Goff, J. Peyronneau, et al., Geophys. Res. Lett. 29 doi:10.1029/2001GL014227 (2002).
S.A. Gilder, M. Le Goff, J.-C. Chervin, et al., Geophys. Res. Lett. 31 doi:10.1029/2004GL019844 (2004).
J.C. Chervin, B. Canny and M. Mancinelli, High Pressure Res. 21 305 (2001).
P. Bertelsen, W. Goetz, M.B. Madsen, et al., Science 305 827 (2004).
Static stress demagnetization
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
547
D.J. Dunlop and J. Arkani-Hamed, J. Geophys. Res. 110 1 (2005).
A.A. Bogdanov and A.Ya. Vlasov, Izv. Phys. Solid Earth 1 24 (1966).
E. Appel and H.C. Soffel, Geophys. Res. Lett. 11 189 (1984).
S. Gilder and M. Le Goff, in Advances in High-Pressure Technology for Geophysical Applications, edited by
J.H. Chen, Y.B. Wang, T.S. Duffy, et al. (Elsevier, Amsterdam, 2005), p. 315.
K.L. Louzada, S.T. Stewart and B.P. Weiss, Proceedings of the Lunar and Planetary Science Meeting XXXVI
(2005).
N. Artemieva, L. Hood and B.A. Ivanov, Geophys. Res. Lett. 32 L22204 (2005).
H.A. Ugalde, N. Artemieva and B. Milkereit, Magnetization on impact structures – Constraints from numerical
modeling and petrophysics, GSA special paper 384: Large Meteorite Impacts III, doi:10.1130/0-8137-2384-1
(2005), p. 25.
P. Rochette, L. Hood, G. Fillion, et al., EOS Trans. 84 561 (2003).
P.S. Mohit, EOS Trans. 85 219 (2004).
P. Rochette, L. Hood, G. Fillion, et al., EOS Trans. 85 219 (2004).